I borrowed that title from Russ Roberts (but inserted the qualifier “mainstream” since, of coure, I love how the Austrians do “macro” topics like business cycle theory). You may recall that Russ was the inspiration for my recent post where I chortled over (what Russ and I thought was) a clearly botched prediction from Krugman that the “fiscal doomsday machine” (Krugman’s term) of the sequester would probably cost 700,000 jobs.
Now in the comments of my post, a few people pointed out that Russ and I (and Scott Sumner, since he has been having fun with the post-sequester economic statistics too) are too quick to claim victory. In particular, surely Krugman didn’t mean that the absolute level of employment would suddenly fall by 700,000 in one month. Rather, Krugman was saying that over the length of the sequester, total employment would eventually be 700,000 lower than it otherwise would have been. Thus there were issues of a counterfactual, and timing, that we would need to consider, before evaluating Krugman’s prediction. As “Joe” pointed out, if we use the Total Nonfarm Employment seasonally adjusted figures from the BLS, and if we choose our baseline of job growth from November 2012 through February 2013 (just before the sequester kicked in), then the average is 247,000 jobs added per month. In contrast, if we look at March 2013 through July 2013 (the latest point available), then the economy under sequestration added only 173,000 jobs per month. So if that discrepancy holds up, it would only take 10 months for Krugman’s prediction to be vindicated.
I absolutely love this sort of thing, where presumably smart guys (Russ, Scott Sumner, and me) can look at the data and walk away with one interpretation, where other presumably smart guys (“Joe” didn’t make any calculation mistakes) can look at the same data and reach a diametrically opposite conclusion. In fact, this is an overriding theme on both my blog and Russ’ posts at Cafe Hayek: There are so many moving parts in macroeconomics that you can ex post fit the observations to just about any theory you want, and you won’t be “lying.” As I like to put it, believing is seeing. That’s why we’re still arguing about the Obama stimulus package, and indeed about the Great Depression. We can’t turn back the clock and re-run the same experiment, this time (say) staying on the gold standard in the 1930s.
My Point Is to Show the Bankruptcy of the Empirical Approach
I know that some Austrians are going to jump on me in the comments, so let me say upfront:
MY POINT IN DWELLING ON THIS STUFF IS TO SHOW THE BANKRUPTCY OF THE EMPIRICAL APPROACH TO ECONOMICS. So for people who say, “Bob, what are you doing?! You’re conceding everything to Krugman when you play the game on his terms!!” I would just say, we have a difference of opinion on the best way to get outsiders to change their minds. Superficially, it sounds entirely reasonable and scientific to take David Friedman’s approach as he laid it out in our debate last summer. It does indeed sound like the Misesians are being weird and medieval if they insist that we can deduce economic laws in an a priori fashion, and no empirical evidence would shake our convictions. Isn’t it the mark of scientific objectivity and humility, to be willing to make falsifiable predictions based on your “model,” and then change your views accordingly when the facts come in?
So, to try to get someone who thinks this way, to understand why it doesn’t work in economics the way it does in physics or chemistry, I think it is entirely worthwhile to show that Krugman et al. aren’t being nearly as “objective” as they think. The Obama stimulus is the best example, of course, where the Romer/Bernstein predictions were hilariously wrong, but no Keynesian batted on eye: “Huh, I guess the economy was worse than we realized, now it’s really good that we ran up the deficit, timid though the stimulus package was.”
Different Baselines Cast “Doomsday Machine” In Different Light
I’ve already relayed “Joe”‘s analysis above, which showed one way that you could vindicate Krugman’s prediction. OK, but Krugman never specified the relevant baseline. As I pointed out to Joe, he conveniently stopped going back in time at the best possible month, because job growth in October 2012 was only 160,000 (below the average under sequester). Is that why Joe started his baseline in November 2012? Only Joe can know for sure.
Let’s pick a less arbitrary baseline. Suppose we look at the 12 months prior to sequester. In that case, using the same government data set that Joe and I were citing in the previous post (and again, I’m not endorsing government stats, I’m just showing how the Keynesians can be wrong even on their own terms), average monthly job growth is 188,000 (compared to the average post-sequester of 173,000). So, if the prior year is the relevant baseline, then Krugman will eventually be right (assuming current trends continued) in 47 months, i.e. in almost four years. Is that what he was trying to say? We’ll come back to that question shortly.
What if we go back two years prior to sequester? In that case, the relevant baseline is average job growth of 163,000/month. Oops, that’s lower than average growth under the sequester, so clearly we can’t go back that far in constructing our baseline.
Finally, my personal favorite, what if we look at average job growth from September 2009–you know, the month after Krugman said that Big Government had saved the economy–until February 2013, when Krugman told us to brace for the fiscal doomsday machine? In that baseline, average job growth during the entire recovery (up to the sequester) was 123,000 jobs per month, a full 50,000 jobs per month less than the growth we’ve seen since the sequester.
Here’s an even more fun fact: In the five months after Krugman declared that Big Government had spared the economy from another Great Depression, the economy lost an average of 131,000 jobs per month. This is in contrast to the average monthly gain of 173,000 jobs, in the five months after Krugman declared that “one of the worst policy ideas in our nation’s history” kicked in.
To be clear, there is no outright contradiction in any of the above. But notice how much Krugman is relying on his Keynesian models, with their counterfactual projections of an alternate universe that we never get to observe, when he so confidently tells us that IS/LM has soared through this crisis with flying colors. A naive examination of the raw data would show that the economy was awful when the stimulus kicked in, and isn’t so bad after the sequester kicked in. The only reason it’s “obvious” to Krugman that the stimulus helped, and the sequester hurt, is that he is using his own Keynesian model to tell him what the world would have been like had there been no stimulus, and no sequester. Fine, but you can’t then use these observations to vindicate the Keynesian model! The model looks vindicated, only if you already thought it was true at the outset. This is why Austrians in the Misesian tradition think economists are fooling themselves when they try to copy the methods of the physicists.
And the Grand Finale: The Macroeconomic Advisers Analysis That Krugman Endorsed Was Totally Wrong
I’ve saved the best for last. All of the above could understandably be construed as quibbling by a Krugman fan. Ah, but what if we click Krugman’s link and actually look at the analysis that generated the “700,000 jobs” figure? Then we’ll see the exact context of that prediction, and realize that it has been repudiated by observations–at least in the same way that the Romer/Bernstein report was totally repudiated.
The Macroeconomic Advisers report first constructs a baseline projection of GDP growth without the sequester, than overlays it with their projections of how GDP growth will be reduced if the sequester happens. This forms the basis of their projections of slower job growth: lower government spending ==> lower output growth ==> fewer workers needed to create that output.
So, if you flip to the tables at the end of the report, you’ll see that its baseline forecast of GDP growth in the second quarter of 2013 was 2.4 percent. But, if the sequester kicks in, in this “Alternative Scenario” they were projecting GDP growth in 2q2013 of only 1.1 percent. Okay, you can see that the sequester was modeled to have a humongous impact on the economy then, in the second quarter of 2013–a reduction of growth of 1.3 percentage points,
almost cutting the baseline growth forecast inby more than half.
Now, leaving the world of the Macroeconomic Advisers Keynesian model, what actually happened in the real world? Well, according to the BEA, the best estimate right now of 2q2013 GDP growth is 2.5 percent.
Does everyone see the absolute deliciousness of this? The actual GDP growth in 2q2013 was higher with the sequester than the Macroeconomic Adviser report said it would be without sequester. It is thus the mirror image of the Romer/Bernstein projection of the stimulus package’s impact on unemployment.
Of course, we can adopt the same trick that the Keynesians used back in 2009. We can now say, “Phew, the economy was better than we realized back in February 2013, and so the awful sequester is not causing as much misery as we had feared. Still, we have slower GDP growth, and less job creation, because of the sequester. It’s still a dumb policy, we just got lucky.”
But at some point, maybe the Keynesians should consider that having government officials spend money, isn’t the path to recovery? Maybe that’s why these “baseline forecasts” have to keep shifting around, in such a way as to justify the Keynesians predictions ex post?
NOTE: If you look at the projections of first quarter GDP growth, then things are better for the Macroeconomic Advisers analysis. But surely a better test of their model is to look at 2q2013, which is all sequester.